Many low-end Android phones could be shut out of Google's new Gemini Intelligence by a minimum RAM requirement. Google has positioned Gemini as the central AI layer of Android 17, but code strings and compatibility notes suggest the assistant may be limited to higher-spec hardware. Reporting from Android Authority and PCMag warn some Android Go devices could lose Assistant without receiving a full Gemini replacement. Google has tied further rollout and compatibility details to its upcoming I/O developer conference.

The read here is simple. Gemini is ambitious, and it will need more than the tiniest phones can offer. Android 17 bundles a deeper, system-level version of Google’s AI called Gemini Intelligence that reaches across apps and user data. The company says it expands on last year’s Magic Cue and adds new capabilities that can act across apps, data, and conversations.

What Gemini brings to Android 17

Gemini’s announced toolset is broad. It includes an agentic Automation feature able to perform multi-step tasks such as booking classes or searching for items from screenshots. There's Rambler, an improved voice-to-text mode that removes filler words while preserving conversational tone. Create My Widget will generate widgets from natural-language prompts, and Auto-Fill will pull context from connected apps to complete forms and tasks more intelligently.

EFTM reported that Google plans granular opt-in controls, so users can enable or disable whole features or specific parts at any time. That opt-in requirement is important for how Gemini handles personal content.

On that front, Personal Intelligence hooks the assistant into Google services including Gmail, Photos, Search, Maps, Calendar, Drive, and YouTube so the model can reference a user’s content to produce tailored responses. Digital Trends reported Personal Intelligence was rolled out to paid Google AI subscribers globally outside the European Economic Area, Switzerland, and the UK, with free users slated to gain access in coming weeks.

Android Headlines documented that Google has already begun making Personal Intelligence available to free users in the United States.

Google has emphasised that Personal Intelligence requires explicit opt-in, and that the assistant references personal content for responses without training the underlying model directly on that content. That distinction addresses privacy concerns while still allowing personalised replies.

Mastheads

Who gets access, and who may lose Assistant

Compatibility is where the promise meets friction. Android Authority has flagged limits that may exclude lower-powered devices. Google previously set minimum system requirements for Gemini, including around 2 GB of RAM. Those requirements matter because many Android Go phones ship with less memory or older processors.

Code strings found in the Assistant Go 2.18.0 update include a deprecation message saying, "Google Assistant is no longer available on this device" and pointing users to check whether Gemini is available for their phone or account type. Android Authority notes there isn't yet a pared-down Gemini equivalent for Android Go hardware, so some low-powered devices that have relied on Assistant Go could face a gap.

Put plainly, Assistant Go appears to be preparing a transition message while Gemini’s ability to run on those devices isn't assured. For owners of very cheap or ageing phones, that could mean losing an assistant they have depended on, without any immediate replacement.

PCMag lays out additional account and device constraints that will shape the real-world rollout. Gemini on phones still depends on some Google Assistant infrastructure for wake phrases and Voice Match. Converting a device to use Gemini requires a personal Google account or an eligible Workspace account. Devices managed under Family Link and Google Workspace for Education accounts for those under 18 are excluded from Gemini access.

PCMag also outlines how users set Gemini as the default assistive app and how to launch it, reinforcing that the user experience will vary by device, region, and account type. That variation matters because two phones running the same Android 17 build could get very different assistant behaviour depending on hardware and the types of Google accounts on the device.

The technical floor that Gemini lays down is a trade-off. A single, capable model across the platform simplifies Google’s roadmap and raises the range of features available on supported devices. But it also concentrates capability among higher-spec hardware, leaving older and budget phones behind unless Google builds a lighter alternative.

For users and phone makers the choices are practical. Manufacturers shipping new mid-range and premium devices will be able to advertise deeper, system-level AI. Makers of entry-level phones must decide whether to boost hardware to meet Gemini’s bar or accept that their devices will continue with a pared-back Assistant or none at all.

App developers and enterprises also need to factor account limits into deployments. Schools and families using managed accounts could be shut out of Gemini, so any product plans that assume system-level AI on every Android device will need guardrails for who can actually use it.

Google hasn't published a full compatibility matrix yet. The company tied these Android 17 announcements to its upcoming I/O developer conference, where it said more details and rollout schedules would be provided.

Related Articles

Google will use its I/O developer conference to confirm timelines and compatibility details for Gemini on Android 17.

This article was created with AI assistance.